I have recently started learning in the field of data science and this explanation of your increased my interest and determination to continue in it. You made the explanation amazing.
Do you mean to say in sentiment analysis, statistics based categorization or mining is not helpful? Could be informative to see a video on that use case.
The video was very clear and precise for me!! Can you please cover more on the tasks involved in text analytics? i.e., Lexical, Syntactical, Semantic, Pragmatic, Discourse analysis?
I think you should cover the ways that text mining for themes using a [keyword type] + [sentiment type] approach can be applied to major nodes in directional graph representations of online discussion. Simple graphing can tell you who is a bot, but applied analysis of the rest allows you to easily profile a node and sometimes identify malicious accounts waging information warfare on behalf of hostile state actors. The information space is a primary attack vector for those who wish to undermine democratic societies.
Depends on the project goals. I would start by defining a dictionary of themes or categories you expect to find in the text. Let\s say the project is food related. One theme could be fried food. "Fried", "battered", "Kentucky", "fish & chips", "onion rings", "tempura", "crispy", "panko" could be some of many key terms to flag a paragraph, comment, or whatever unit of partition as involving fried food. From there, you could further divide entries flagged as fried into subcategories of good or bad. First you use an easy general classifer. Words like "disgusting" or "nasty" would automatically be flagged as negative connotation, while terms like "tasty" or "mouth-watering" would be flagged as good. The best part is that this general good/bad keyword set is applicable to all your other food types. But even further, we could make a fried.sentiment keyword set specifically built to pick up anything we may have missed. "greasy" could be neutral, so in fried.sentiment we would have "too greasy" as a negative flag but "greasy goodness" or "nice and greasy" as a positive flag. You could event assign a scoring mechanism for large documents so that the total number of good/bad flags is tallied. Only when the number of good and bad flags is nearly even would you have to take the time to line by line examine the particular doc.
He's so cute and looks like he's trying to hold back a laugh the entire time :) I could listen to him talk for hours
Brilliant video - so well explained and really engaging to watch. A great way to supplement my learning :)
I have recently started learning in the field of data science and this explanation of your increased my interest and determination to continue in it. You made the explanation amazing.
I have a presentation tomorrow on text mining. This really helped me.
Do you mean to say in sentiment analysis, statistics based categorization or mining is not helpful? Could be informative to see a video on that use case.
Nicely explained!
Thank you very much for great clarity of concept and neat presentation !! 🙏😊
The video was very clear and precise for me!!
Can you please cover more on the tasks involved in text analytics? i.e., Lexical, Syntactical, Semantic, Pragmatic, Discourse analysis?
Nice overview!
Thank you I enjoyed and had fun how you explained it
I would like to know a practical case of use about text mining in the industry (maintenance area)
I liked it
What tools are out there that i would be able to try text mining?
This is not crypto
Python or maybe R. Check out nltk first.
I think you should cover the ways that text mining for themes using a [keyword type] + [sentiment type] approach can be applied to major nodes in directional graph representations of online discussion. Simple graphing can tell you who is a bot, but applied analysis of the rest allows you to easily profile a node and sometimes identify malicious accounts waging information warfare on behalf of hostile state actors. The information space is a primary attack vector for those who wish to undermine democratic societies.
brilliant thanks
Sir can you share some information abt Mobile Analytics in upcoming video
Great, now how can I apply this to a body of text totaling 2 million words? Right across 900 + files, all geared towards one project?
Depends on the project goals. I would start by defining a dictionary of themes or categories you expect to find in the text. Let\s say the project is food related. One theme could be fried food. "Fried", "battered", "Kentucky", "fish & chips", "onion rings", "tempura", "crispy", "panko" could be some of many key terms to flag a paragraph, comment, or whatever unit of partition as involving fried food. From there, you could further divide entries flagged as fried into subcategories of good or bad. First you use an easy general classifer. Words like "disgusting" or "nasty" would automatically be flagged as negative connotation, while terms like "tasty" or "mouth-watering" would be flagged as good. The best part is that this general good/bad keyword set is applicable to all your other food types. But even further, we could make a fried.sentiment keyword set specifically built to pick up anything we may have missed. "greasy" could be neutral, so in fried.sentiment we would have "too greasy" as a negative flag but "greasy goodness" or "nice and greasy" as a positive flag. You could event assign a scoring mechanism for large documents so that the total number of good/bad flags is tallied. Only when the number of good and bad flags is nearly even would you have to take the time to line by line examine the particular doc.
I want to learn how to mirror writing
Search on "lightboard video".
Brilliant ASMR
How can this bro write so good on the glassboard 🙃
See ibm.biz/write-backwards
I do not believe the shirt story is real I think he made it up to fit with the video
Hah. Of course:)